So let's take a look at some of the visual cloud use cases in a little more detail. So we're going to break this up into some segmentation. So we've got five of them laid out here. And while they may look like they're certainly distinct, there maybe a little bit of shadowing that takes place from one to the other. And that's going to be OK. So the media processing and delivery use case. So this is where we were talking about the encoding and decoding and transcoding-- the streaming of video content either from a public or private source into the cloud itself. Our next large segmentation are cloud graphics. So that's where we find the remote desktop and rendering inside the cloud itself so that you can think about creation of content interactively rather than sourcing content from the media delivery. Cloud gaming-- we can't emphasize the emergence possibility of this one enough. With the media encoding that takes place there and to deliver that game content, not only to the consumers of the game, but potentially also to an audience. This is becoming-- e-gaming and e-sports are becoming a large marketing opportunity. And then, inside the network itself, the media analytics that can go on adding intelligence to the media streams. Whether that's artificial intelligence guiding or through some type of machine learning. And then, potentially also offline media analytics to enhance media usage for smart city applications or for security applications, for example, that you might find associated with some type of a venue. And then, we've got the immersive media. And this is where we find the AR and VR to provide a fluid user experience. Some of the things there may be 3D processing or gaming so that you can get an any-seat-anywhere type of an experience from a venue, whether that's sports or a concert or some type of spectator activity. So these are the very large segmentations. And we're going to go into a couple of these a little bit more. So we talked about the smart city activity and video surveillance or security as a significant portion here. You may even be able to extend this into applications at the enterprise and at the home. Certainly, we're aware of what we can do now with video cameras and access in the home. But that traffic, in some cases, remains localized. But if we enable it to go through the access network and it can operate on it, we open up opportunities in a variety of other industries-- whether that's in transportation for vehicle identification, plate recognition, lane change education, congestion control. If you think you've got something wrong-- if you see abnormal lane changes going on. Maybe you need to dispatch a truck because something has fallen-- a tire in the middle of the road. And you're getting abnormal activity from that type. Traffic management for jam congestion or moving slowly is also an area that we might find advantageous once we've enabled the system. Facial recognition-- obviously, we're talking now here about security implications and everything that goes into there. Moving object identification-- again, being able to identify particular objects. Whether this is in a factory and we're doing some type of inventory sorting as things move down through that door. Behavioral analytics are also all capable once we begin sourcing this wealth of information and then driving this traffic through into the network. Another interesting application use case is-- most of the devices that we carry around in our pocket are one of two operating systems. And there's some energy going on for one of those operating systems to enable the Android edge cloud for remote rendering. So here, you might think about a user who is using their mobile handset on that environment and need more resources for whatever the activity is. Maybe they are interacting with some type of streaming activity or gaming. And the visual cloud may be able to provide application resources that bring additional compute resources that are not in that handset to that device, as a fine example. The media analytics that we mentioned before. This is an area for the content delivery networks. So again, a lot of the network utilization from the core of the network is for media streaming today. Some of that's consumed by the handset over the wireless network. An awful lot of that is consumed at the consumer premises during the evening hours. We're watching videos or streaming YouTube or whether we're watching a long-content type of a movie. But inside the network itself, to relieve that congestion, we're seeing the emergence of content delivery at the edge. And that allows the network to change the time slice, if you will. Not significantly, but a little bit. If you're streaming a five-minute video and there's going to be a certain number of hundreds of packets, the network may be able to detect that and decide to queue up that video at those edge locations when there's less congestion in the network. And we're talking second-by-second type of options that the network can make those decisions on to move that content actually from the source and queue it up preemptively so that when the HTTP GET goes out from that set-top box, it doesn't have to go all the way deep into the network. But the CDN at the edge will be able to service that. In addition, when we look at the analytics that go into place, we can do localized ad insertion then. You may know that there's consumers in a particular neighborhood who just experienced some storm issues. And you might want to have home repair ads that are localized to them as an option that comes in, for example. There are a variety of other things that we can do from an analytics standpoint. Once we recognize the fact that we can draw those compute resources closer to the consumer and apply analytics, it helps identify needs or opportunities for that consumer. The media content creation and delivery use cases are rather extensive. So we've really got three areas in here. That is the broadcasting of live or the video on demand. And here, you can think about live sports-type activities or your town hall meetings, the c-span type of streaming where it really is instantaneous. But it still needs to be delivered through the network. And you've still got some of those transcoding opportunities that can take place in that. And in addition, you're delivering it in multiple formats. Again, the consumer may be consuming on an HDR 4K type of a consumption model. Or they may be consuming it on a small device. And in addition, that consumer may in fact choose to flick it from one device to the other depending on their mobility or their location. And you want that network to be able to adapt to that seamlessly for that user experience. And then through the network, or Over The Top, or the OTT, falls into, again, two categories. The video on demand, which we understand. And we know what those are. Certainly, the cloud media processes that are possible out there. Again, to relieve the potential for congestion on the network if you have a high-demand activity that's going on-- breaking news type of an activity-- that's a great model for that in the live or the delivery of that in the over the top. So those are the types of areas that we certainly see a significant opportunity for their growth. We mentioned a little bit of the media analytics. So that also goes then into not only the opportunity for monetization from the comm service provider. But if we think about security implications, there are visual search and screening applications-- smart city and surveillance, transportation and robotic vision. Again, from an industrialization standpoint or from transportation. You think about the distribution of freight and goods and the video analytics. If we've got the ability to sense the location of that freight or that transfer are great applications that we see. And then finally, the video monitoring. So you may be able to provide enhanced services to dispatch resources when necessary based on analytics that you get based on video monitoring. The immersive video-- so again, when we think about the immersive video, we think AR and VR and 360-type viewing. This is the area where that compute resource really needs to be close to the end consumer. Because that latency of delay that takes place can cause biological, physiological problems with that delay. We really need to see compute resources in that area that have turnaround times of less than five milliseconds to prevent us from experience adversely the latency from those areas. So the growth drivers there are still in their very early stages. But we do see that, as we mentioned, in gaming and immersive sports entertainment. And chances are, it's the next generation that is going to drive this growth, as they begin to just accept that's the way that we experience the world around us-- interact with those types of things in our leisure time. So the cloud graphics use cases-- again, we talked a little bit about cloud gaming. Very interesting workloads that can come into play there. So you've got a large group of interactive people that are either local and enjoy some type of e-sports type of activity, or distributed doing some leisure time game gaming activities. A lot of rendering, a lot of transfer that takes place in that type of an environment. Whether it's ray tracing to recreate the images to give a better 3D representation so that the users feel more immersed in the game. Or whether this is some type of remote desktop operation where you've got someone who's creating content. And they need a significant amount of compute resources because of the animation that they're trying to generate. And you really don't want to put those large computer loads at every one of the locations that you may have your content creators at. But you want to be able to localize that so that you can distribute or share that compute resource. And then, still get a very quick turnaround so that your artist-- your creator-- of this content is not waiting for the compute resources in order for them to continue with their creative process.
